Feedback-Dependence of Correlated Firing in Globally Coupled Networks

It is known that, in feed-forward nets, the degree of neural correlation generally increases with firing rate. Here, we study the correlations of neurons that are part of a homogeneous global feedback network, under the influence of partially correlated external input. By using numerical simulations of a network of noisy leaky integrate-and-fire neurons with delayed and smoothed spike-driven feedback, we obtain a non-monotonic relationship between the correlation coefficient and the strength of inhibitory feedback connections. This non-monotonic relationship can be explained by the interplay between the mean rate and the regularity of firing activity caused by the inhibitory feedback connections. We also show that this non-monotonic relationship is robust in both sub-threshold and supra-threshold dynamic regimes, for low and moderate internal noise levels, as well as when the network is heterogeneous. Our results point to a potent functional role for feedback as a modulator of correlated activity in neural networks.